7.1 INTRODUCTION

The test methods discussed in previous chapters have a common property; i.e., the sample sizes are predetermined before the test. In this chapter, another method, the Sequential Probability Ratio Test (SPRT) [1] method, is presented, which has the advantage that in general the average sample size needed to reach a decision is smaller than that for the previous test plans.

An ordinary accept-reject test plan consists of a pair of two numbers ( n, c), where n is the sample size and c is the reject number of failures. For example, assume that the required minimum MTBF is m _L. An ( n, c) plan is used to decide if the lot of an equipment should be accepted or rejected. The test is carried out as follows: A random sample of n equipment is drawn from the lot and all n are put to a life test simultaneously for the duration t _d. If r failures are observed during the test, then the lot is accepted if

r < c;

otherwise, it is rejected.

In the test plan given previously, the average sample size is n. Test plans in which the sample size cannot be predetermined and each time one sample is drawn from the lot and put in the test, after each test, the accumulated test results are compared with predetermined decision rules to conclude whether to (1) accept the lot, (2) reject the...